Blind Image Decomposition

Blind image decomposition aims to separate an image into its constituent components, such as clean image content and various degradations (e.g., blur, noise, atmospheric effects), enabling targeted restoration or manipulation. Recent research emphasizes developing sophisticated models, often based on diffusion processes or U-Net architectures, to achieve accurate component estimation and, increasingly, user-control over which degradations are removed or retained. This capability is crucial for improving image quality in challenging conditions and offers valuable tools for image editing and copyright management, impacting fields like remote sensing, medical imaging, and digital forensics.